232 A Case of Abnormal Trichromatic Colour Vision due to a in the Spectrum of the Green-Sensation Curve .
By Sir W. de W. Abney , K.C.B. , F.R.S. , and W. Watson , D.Sc .
, F.R.S. ( Received June 13 , \#151 ; Read June 26 , 1913 .
) The authors have each separately dealt with the question of complete and incomplete colour-blindness caused by the absence of , or decrease in , the response to stimulation of the red or green perceiving apparatus which is functional in the case of vision .
We have shown that a large number of cases of defective perception of colour are simply explained on this hypothesis .
Abnormality of colour vision may also be due to a shift in , or an alteration in form of , one of the sensation curves .
In the present paper we discuss the effect of one type of shift on the colour perception and give the results of a series of measurements which show that such a shift , without any alteration of form , does sometimes occur .
In a paper which appeared in the ' Proceedings of the Royal Society , ' * one of us indicated how a shift , that is a displacement of the whole curve so that the maximum is displaced to a different wave-length , of one of the sensation curves could be detected by a simple spectrum test , which is as follows .
When yellow light of wave-length 5760 A.U. is mixed with blue light of wave-length 4610 in suitable proportions the mixed light looks to the normal eye exactly the same as the white light from the crater of the electric arc both in hue and brightness .
If a person who has a shift in one of the sensation curves is shown this match it will not appear correct to him nor can it be made correct by any alteration in the proportions of the yellow and blue lights .
There will , however , in every case be found a position of the slit through which the yellow light is obtained with which a satisfactory match can be obtained .
In other words the wave-length of the light which is complementary to the blue will be different to that of normal vision .
On the other hand , where the colour perception defect is due to a deficiency of one of the sensations the observer will agree that the match ' made by the normal is correct , although in most cases he will also consider correct matches made when the yellow slit is moved to one or other side of the normal setting , the range being the larger the greater the defect .
In order to see why in the case of colour defect due to a deficiency in one of the sensations the normal match appears correct it is necessary to * 'Roy .
Soc. Pioc .
, ' 1912 , A , vol. 87 .
A Case of Abnormal Trichromatic Colour Vision .
233 consider the sensation curves .
In fig. 1 are given the sensation curves of the spectrum colours for the normal eve , the source of light being the crater of an arc with a horizontal positive carbon , i.e. one in which the crater directly faces the slit of the spectroscope .
The sensations are given in terms of luminosity* so that the sum of the ordinates of the three curves for any wave-length ZZ 24 26 28 30 32 34 36 38 40 42 44 46 50 52 54 56 58 60 62 2000AJJ 5,000 5500 6,000 6500 Fig. 1.\#151 ; Scale of prismatic spectrum ( S.S.N. ) , showing wave-lengths .
is equal to the luminosity at that wave-length .
Since the luminosity of the blue sensation is so small compared to the others the blue sensation curve has been plotted on a scale 100 times as great as that adopted for the red and green sensations .
When considering the matches made between mixtures of different coloured lights it is , however , often convenient to adopt a different scale , namely , one in which the areas of the three sensation curves are all equal .
With this scale equal ordinates of the three sensation curves correspond to a mixture which will appear to the normal eye as white .
The areas of the three sensation curves on the luminosity scale are Eed = 579 , Green = 248 and Blue = 3'26 , and these correspond to the sensation produced by the whole combined spectrum , that is by white light .
If we multiply the ordinates of the green sensation curve by 579/ 248 or 2*21 , and those of the blue sensation curve by 579/ 3-26 or 117 , the green and blue sensation curves will have the same area as the red sensation curve , the three sensation curves plotted on this scale are shown by the full line curves in fig. 2 .
An explanation of what is meant by the luminosity of a coloured light is given on p. 404 of vol. 88 ( A , 1913 ) of the ' Proceedings .
' U 2 Sir W. de W. Abney and Dr. W. Watson .
Let ns now consider an observer who has , say , only half the normal green sensation , so that on the luminosity scale the ordinates of his green-sensation curve will he half those for the normal curve .
On the equal area scale , however , his green-sensation curve will be the same as the normal , for since ioo $ 20 \lt ; D H 60 3 50 o 40 30 *o ri2 20 m io 1 y s ; A 7 \ \ A ( / 4r l t A \amp ; \ \ u. \ d/ / / / \ V \ / / ,7 A \ A \ \ A \lt ; 4^00 Fig. 2.I I t i i i ] i 4500 5,000 -Scale of prismatic spectrum ( S.S.N. ) , showing wave-lengths .
i pi i i i | i 6,000 6,500 ^oooAU .
the area of the green-sensation curve on the luminosity scale is now 124 , to obtain the equal area scale we must multiply by 579/ 124 or 4*42 .
Hence , since the multiplier is twice as great as for the normal , the resulting curve will be the same as for the normal .
How if we take a yellow at S.S.N.* 48*4 ( 5760 A.U. ) , at a ( fig. 2 ) , and mix it with a violet at S.S.H. 9 5 ( 4235 A.U. ) , at b ( fig. 2 ) , the width of the violet slit being 2*5 times that of the yellow , and read off the three sensations at these places from the full line curves given in fig. 2 , the values in the violet being multiplied by 2*5 , we get the following numbers:\#151 ; * The values of the scale numbers given in the diagrams in terms of wave-length are as follows:\#151 ; S.S.N. Wave-length .
o A.U. S.S.N. Wave-length .
0 A.U. 62 6957 34 5002 60 6728 32 4924 58 6521 30 4848 56 6330 28 4776 54 6152 26 4707 52 5996 24 4639 50 5850 22 4578 48 5720 20 4517 46 5596 18 4459 44 5481 16 4404 42 5373 14 4349 40 5270 12 4296 38 5172 10 4245 36 5085 8 4198 A Case of Abnormal Trichromatic Colour Vision .
Table I. Position of slit .
Sensations .
Red .
Green .
Blue .
a 67 *9 69 -3 1 -6 b 1 -4 0 67 -7 Sums 69 3 69 -3 69-3 Since the sums for the three sensations are the same it follows that the mixture will look white to the normal eye .
Further , since the curves for the person who has half the green sensation are precisely similar , the sums will be equal for him also and hence he will match the mixed colour with his own white .
Similarly for any other case of colour defect , where the defect is due to a deficiency of one of the sensations , the equal areas of the sensations will be the same as those of the normal eye .
Although the person who has a defect of one of the sensations will agree with the normal match , it will be found that when making the match the position of the yellow slit can be moved some little distance from the correct position for the normal without the match becoming defective to him .
This want of definiteness of the mixture required for a match when one of the colour sensations is defective can be illustrated by fatiguing the normal eye ( say ) with red so that temporarily the red sensation is defective .
This effect is illustrated in the following matches made by one of us .
In each case white was matched by a mixture of red , green and violet light by altering the width of the slits through which the coloured light proceeded .
In the case of both the unfatigued and the fatigued eye not only was a correct match made but also one when the green slit was so much reduced that the deficiency in green was just observable and one where this slit was so much opened that the excess of green was just perceptible , the red slit being kept at a constant width throughout .
The numbers obtained for the width of the green slit are given in Table II .
The range between a noticeable excess and defect of green is for the unfatigued eye 4-5 , and for the fatigued eye 16-2 , and a match correct for the unfatigued eye is correct for the fatigued eye .
If the sensation curves are the same for a given observer as for the normal , except that one of them is shifted along the spectrum , quite a different result will be obtained .
Thus suppose that the green-sensation curve is shifted towards the red end of the spectrum by an amount equal to 236 Sir W. de W. Abney and Dr. W. Watson .
Table II .
Character of match .
Width of green slit .
Unfatigued .
Fatigued .
Correct 22 -7 23 -5 Too little green 20 5 11 -o Too much green 25 -0 27 -2 2 S.S.N. and occupies the position shown by the dotted curves in figs. 1 and 2 .
The sensations at the points a and b for such a person are as follows:\#151 ; Table III .
Position of slit .
Sensations .
Red .
Green .
Blue .
a b 67 -9 1-4 79 -2 0 1-6 67-7 Sums 69 -3 79 -2 1 69 -3 The sums of the blue and red sensations are still equal but the sum of the green sensation ordinates is greater , and hence the mixed colour will not match the white but will appear to such a person too green .
If , however , the yellow slit is moved towards the red to c , fig. 2 ( S.SJST .
49'8 , 5860 A..U .
) and the width of the violet slit is made 272 times that of the yellow slit we get Table IV .
Sensations .
Position of slit .
Bed .
Green .
Blue .
c 71 -8 73 -5 0 5 b 1 -7 0 73 0 Sums 73 -5 73 5 73 o The three sums are now equal and hence the mixed colour will match the white to such an observer , although it will appear orange to normal vision .
A Case of Abnormal Trichromatic Colour Vision .
237 Hence , if an observer does not agree with the normal when violet and yellow are matched to form white but requires that the yellow slit be moved towards the red to form a match , we conclude that his green-sensation curve is displaced towards the red and vice versa .
A movement of the red-sensation Qurve would produce a similar change but since we have not hitherto had an opportunity of finding such a case we shall in this paper only consider the effect of a shift of the green-sensation curve .
It may , however , be pointed out that a shift of the red-sensation curve would be accompanied by an increase or decrease in the luminosity of the extreme red end of the spectrum according as the shift is away from or towards the green .
As far as the shift of the green-sensation curve is concerned we have hitherto only met with displacements towards the red end of the spectrum .
In the case of a shift of the green-sensation towards the red amounting to 2 S.S.H. as indicated by the dotted curves in figs. 1 and 2 we should expect the following effects to be produced :\#151 ; 1 .
The part of the spectrum which to the normal appears yellow will appear greenish , for owing to the displacement the green sensation excited will be greater than in the normal .
In the same way what appears orange to the normal will appear yellow , and so on .
2 .
If we place three slits in the spectrum , one at the place of the red lithium line , d , fig. 2 , another at the b magnesium line e , fig. 2 , and the third in the violet at b , fig. 2 , and by varying the width of the three slits produce a mixture which to the normal appears to match the white , this match will not appear correct to the observer with the green shift ( whom , for short , we may designate by 0 .
, the normal being indicated by N. ) .
To 0 .
the match will be imperfect , for the green sensation he receives from the light passing through the green slit at e will not be as great as it is to 1ST .
By opening the green slit we can , however , obtain a match which is correct for 0 .
, but his match will appear green to N. , and he will never agree that the normal match is correct .
If now we move the green slit to f , fig. 2 , where the normal green-sensation curve cuts the displaced curve we shall find that a match which is correct for N. is also correct for O. , and that either can detect a small departure from this setting.* 3 .
The above is one arrangement of the three slits such that 0 .
and 1ST .
make the same match .
Another such position is obtained if the red slit is moved to g , fig. 2 ( S.S.H. 52'4 , 6000 JLU .
) .
The light which now comes through the red slit excites green sensation in the case of both N. and 0 .
, but * We have neglected any small variation in the amount of violet which may be necessary owing to small differences of macular pigmentation .
Sir W. de W. Abney and Dr. W. Watson .
to a greater extent in the case of the latter .
Since the red and blue sensations are the same for both it will be sufficient to consider the equality of the red and green in the two cases .
If the width of the red slit is 06 2 times that of the green we get the following values of the sensations on the equal area scales:\#151 ; Table Y. Sensations .
Position of slit .
N. 0 .
Red .
Green .
Red .
Green .
e 21 -1 48 -5 21 -1 34 -5 9 43 -8 16 *4 43-8 30 -4 Sums 64 -9 64 -9 64 -9 64 -9 where the sums are the same in the two cases and hence the mixture appears white to both 1ST .
and 0 .
Thus if the red and green slits are kept in a constant position and the red slit is gradually moved up towards the green the matches made by 0 .
appear to K at first too green , but the excess of green gradually decreases till the red slit is at S.S.N. 52*4 .
If the red slit is moved further towards the green the mixture which appears correct to 0 .
will then appear too red to N. 4 .
Owing to the displacement of the green-sensation curve O. 's luminosity curve will be higher than the normal on the red side of the point where the normal and displaced green-sensation luminosity curves cut and lower on the green side of this point , for the ordinates of the luminosity curve are the sums of the ordinates of the three luminosity sensation curves .
The resulting luminosity curve for a displacement of 2 S.S.N. towards the red is shown in fig. 1 by the dotted curve , the corresponding normal curve being given by the thick continuous line .
Up to quite recently those observers who gave indications of a shift of the sensation curves have not been able to devote the time necessary for a full investigation .
We have , however , been fortunate enough to have in the Senior Physics Class at the Eoyal College of Science this year a case of such a shift , and this gentleman ( R. ) has been good enough to devote the necessary time to carrying out a complete investigation of his colour sensations .
He has proved to be a most accurate observer , and we can place complete reliance on his observations .
He made a series of matches throughout the spectrum A Case of Abnormal Trichromatic Colour Vision .
239 from which his sensation curves have been deduced in the manner described by one of us in a previous paper.* His red and practically his blue sensation curves are identical with those of the normal , but his green-sensation curve is markedly different .
The numbers obtained for the green sensation are shown by the crosses in fig. 3 .
It will be observed that E/ s green-sensation curve is 'O 5/ )oo 5,500 6,000 6 , Fig. 3.\#151 ; Scale of prismatic spectrum ( S.S.N. ) , showing wave-lengths .
similar in shape to the normal , which is shown by the continuous line on the figure , but that it is displaced by about 2 S.S.N. towards the red end of the spectrum .
Thus to him the maximum for the electric arc light occurs at wave-length 5690 A.U. in place of at 5575 JLlL , which is that of the normal .
When white is matched by mixing light which passes three slits placed at the points d , e , and b , fig. 2 ( 6705 , 5190 , 4235 A.U. ) , E. requires very much more green than the normal , as is shown by the first line in Table VI .
If , however , the red slit is moved towards the yellow , the green and violet slits remaining fixed in position , the excess of green required by E. got less and less , till finally a position for the red slit was found where the mixture matched white both to E. and to the normal .
If the red slit is moved further towards the yellow E. required less green than the normal , so that his mixture looked slightly red to the normal .
The above changes indicated that in the case of E. we had to deal with a shift of the green-sensation curve .
The changes in the slit widths required to match white both by E. and by a normal ( W. W. ) are shown in Table VI .
The positions of the green and violet slits were kept constant , as w:as also the width of the green slit , the * ' Phil. Trans. , ' 1905 , A , vol. 205 , p. 333 .
Sir W. de W. Abney and Dr. W. Watson .
match being obtained by varying the widths of the red and violet slits and the brightness of the comparison white .
With the red slit at 52'5 the match made by either was correct for the other .
Table YI .
Position of " red " slit .
Slit widths .
R. W. W. S.S.N. A.U. Red .
Green .
Violet .
Red .
Green .
Yiolet .
59 '8 6710 28 -3 17 '2 26-0 62 *5 17 -2 26 -0 57 '1 6450 8-5 17 *2 26 -0 16 -5 17 -2 26 -0 54 -5 6200 7-2 17*2 26 -0 8-6 17 *2 26 -0 53 -0 6080 9*2 17 -2 33 -0 9-2 17 -2 33 -0 52 -5 6040 10 -4 17 *2 33 -0 9-5 17 -2 33 -0 52 -0 6000 14 -8 17 -2 45 -0 10 -o 17 -2 45 -0 When matching D-light with a mixture of red and green light , if the red is at the red lithium line , R. required considerably more green than did the normal .
If , however , the red slit were moved towards the yellow , just as in the case of the white matches , the excess of green gradually decreased , though , owing to the fact that the D-light in the case of R. excites the green sensation more strongly than the normal , we did not get the marked change in the appearance of R. 's match to the normal which has been referred to in the case of the white match .
For this reason the white match is preferable to the Rayleigh match for bringing out the characteristic changes when the position of the red slit is altered .
Another advantage of the white match is that the yellow produced by mixing green and red to match the D-light is less saturated than the D-light itself , and this causes considerable difficulty with some observers when making the match .
When matching white by a mixture of violet light ( 9'5 S.S.bL , 4235 A.U. ) and yellow light the following results were obtained:\#151 ; W. W. R. Position of yellow slit ... 48'9 S.S.N. or 5780 A.U. 50'0 S.S.N , or 5860 A.U. showing that the complementary to the violet is in the case of R. displaced towards the red , as has been shown on p. 237 , we should obtain if the green-sensation curve was shifted towards the red end of the spectrum .
Again it was found that if the red and violet slits were in the standard positions d and b , fig. 2 , a position for the green slit , / , fig. 2 , could be found such that the mixed light matched white both for R. and for N. A Case of Abnormal Trichromatic Colour Vision .
241 The luminosity curve obtained by R. , using the equality of brightness method , is given by the upper dotted curve in fig. 3 , the continuous line being the normal curve .
In fig. 4 are given the points corresponding to four sets of luminosity measurements made by the flicker method , the crosses giving 26 28 50 32 34 36 38 40 42 44 46 48 50 52 54 56 58 60 62 1 1 1 | 1 1 11 | 1 1 \lt ; 1 I .
1 1 1 1 I 1 1 1 1 | 5poo 5,500 6,000 6500 SO Fig. 4.\#151 ; Scale of prismatic spectrum , showing wave-lengths .
the values obtained by R. and the circles those obtained by one of the authors ( W. W. ) .
It will be observed that R. 's luminosity curve agrees with what it ought to be if his green-sensation curve is displaced towards the green by about 2 S.S.N. , such a calculated curve being given in fig. 1 .
Attention may be drawn to the fact that R. 's luminosity curve does not agree with the normal at S.S.H. 48'6 , as it would if R. 's abnormality were due to a deficiency of either the red or green sensations as has been shown in a recent paper by one of us.* It will thus be seen that the results obtained by R. agree in all respects with what we should expect if there is a shift of the green-sensation curve and which are given on p. 237 , so that they form a strong support not only of the green-sensation curve obtained by R. but also of the normal curves published by one of the authors .
Further , it is difficult to see how the various matches made by R. , which are simply and consistently explained on the trichromatic theory , can be explained on any of the other commonly held theories of colour vision .
* ' Roy .
Soc. Proc. , ' 1913 , A , vol. 88 , p. 404 .
242 Sir W. de W. Abney and Dr. W. Watson .
It is of practical importance to consider what effect a shift of the green-sensation curve such as that exhibited by It .
will have on the power of discriminating colours , particularly those colours which are used as signals at sea and on railways .
As has been mentioned , one effect of the displacement of the green-sensation curve is that the part of the spectrum which to the normal appears yellow , to such persons appears green or greenish .
Thus E. places the change from green to yellow in the spectrum at 49-9 S.S.N. or 5810 A.U. while to a normal ( W. W. ) this point appeared to be at 48-8 S.S.N. or 5780 A.U. One effect of this difference is that a light , such as that given by a paraffin lamp , which to the normal appears decidedly yellow appears to a person with the shift of a greenish hue and in fact E. often calls such a light green .
Another effect of the displacement is that the perception of a green light when diluted with white light is very much more difficult than for the normal .
The reason for this effect is at once apparent from a consideration of the sensation curves .
Consider a green at S.S.N. 36 ( 5090 A.U. ) .
At this point in the spectrum the red and blue sensation curves for N. on the equal area scales intersect .
Hence we may regard the effect produced by light of this wave-length as an amount of green sensation ( represented by the difference between the ordinate of the green-sensation curve and the ordinate of either the red or the blue-sensation curves ) diluted by white light ( this white light corresponding to the equal amounts of red , green , and blue sensation excited ) .
It will be observed that the amount of the diluting white is the same for the normal and the person with the displaced green curve ; but to 0 .
the amount of residual green sensation is less than half that of the normal .
In other words the green perceived by E when light of this wavelength enters his eye is very much more diluted than it is to a person having normal colour vision .
As a certain amount of dilution with white light will obliterate the perception of green in the coloured ray , it follows that to O. the amount of white light which will obliterate it is considerably less for O. than it is for a normal vision .
This effect is clearly indicated by some measurements made by E. that are shown in fig. 5 .
The experiment consisted in determining the amount of any given coloured light which could be added to a given amount of white light so that in the resulting mixture the colour was only just recognisable .
The ordinates in fig. 5 give the luminosity of the coloured light expressed as a percentage of the luminosity of the mixture , the crosses being the values obtained by E. and the circles those obtained by W. W. It will be observed that in the red the two agree as they also do in the blue .
In the green , however , E. requires more than twice as much green light to be mixed with the white than does a person having normal colour vision .
A Case of Abnormal Trichromatic Colour Vision .
243 To see whether the above given explanation is adequate to explain the effect observed in the case of E. we may calculate the proportion the residual green sensation bears to the total white , that is the added white plus the white due to the equal stimulation of the three sensations for the mixture of green at S.S.N. 35 .
The results are given in the following table:\#151 ; Luminosity of colour when just This colour consists of\#151 ; Total Ratio of residual colour to total white .
perceptible ( white = 100 ) .
White .
| Residual colour .
white .
R 27*7 24 '1 3 '6 124 *1 106 '6 0 029 0*031 w. w 9'9 6'6 3'3 It will be observed that the ratio of colour to white when the colour is just observable , when allowance is made for the want of saturation of the colour sensation is the same for E. as for the normal .
Another effect produced by this want of saturation is observed if the Fig. 5 .
Fig. 6 .
minimum brightness is determined at which a 'coloured light can be distinguished from a white light of the same intensity placed alongside , in other words , when the chromatic threshold is determined .
In fig. 6 are given the illuminations on a screen ( 3'2 x 1'6 cm .
) placed at a distance of 66 cm .
from the eye when the coloured light looks indistinguishable from a similar patch of white light alongside .
In the red and the blue E. and W. W. agree , but in the green there is a marked difference , E/ s threshold for colour being markedly higher than the normal .
The above results both as to the effect of dilution with white and as to 244 Sir W. de W. Abney and Dr. W. Watson .
the chromatic threshold are of great practical importance , for they both affect the power of an observer to identify green lights such as those used at sea .
These lights are never pure spectral lights , though they are spectral colours diluted with white .
Thus the Board of Trade standard light-green light can be matched by a mixture of spectral green at S.S.N. 37'4 or 5115 A.U. , with an equal amount of white ( arc light ) .
Further , if the size of the image of the coloured patch on the retina is diminished , it must be remembered that the amount of white required to extinguish a spectral colour is very much reduced .
The increased want of saturation of the green produced by a displacement of the green-sensation curve towards the red is clearly brought out by the curves given in fig. 7 .
In this figure the spectrum is represented in terms Fig. 7.\#151 ; Newton 's colour diagram of the arc-light spectrum .
The three colour sensations which make white light being shown as of equal value .
of sensations on a triangular Newton 's colour diagram .
The corners of the triangle represent the red , green and blue sensations , the scale adopted being the equal area scale used in fig. 2 , so that the centre W of the triangle represents white .
The amount of either of the sensations corresponding to A Case of Abnormal Trichromatic Colour Vision .
245 any point within the triangle is inversely proportional to the distance of this point from the corner corresponding to that sensation .
The spectrum , as it appears to the normal eye , is represented by the continuous line , the numbers written alongside the line representing the different parts of the spectrum expressed in S.S.K The dotted line gives the spectrum as it appears to a person who has the green-sensation curve displaced by 2 S.S.N. towards the red .
Even to the normal eye the green part of the spectrum corresponds to a green sensation diluted by much white , for the curve representing the spectrum passes about half-way between the green-sensation corner and the white point TV .
In the case of the green-sensation shift the curve passes much nearer the white point W , indicating that the colour seen is less saturated than it is to the normal eye .
If the green-sensation curve were shifted by 3'6 S.S.IST .
towards the red , the green-sensation curve on the equal area scale ( fig. 2 ) would pass through the point where the red and blue sensation curves intersect , and hence light of the wave-length ( 5090 A.U. ) corresponding to this point would produce the sensation of white to such a person .
In other words , there would be a neutral point in his spectrum .
In such a case the curve representing the spectrum on the Newton diagram would pass through the point W. As a result of our study of the case of R. , as well as of some more isolated observations made on others , we consider that it is extremely probable that the particular class of colour abnormality which was first investigated by Lord Rayleigh by means of his instrument for matching D-light by a mixture of red and green light , and which is generally referred to as anomalous trichromatism , is due to a shift of the sensation curves , and that green anomalous trichromates are persons who have the green-sensation curve displaced towards the red end of the spectrum .
It must , however , be remembered that abnormal matches may be obtained by persons who in place of a shift have a reduced sensation for red or green .
The two classes are , however , distinguished , as has been pointed out earlier , by the fact that in the case of a deficient sensation the accuracy with which matches can be made is less than in the case of either the normal eye or one in which one of the sensation curves is shifted .
In conclusion , the authors would like to express their obligation to Mr. R. , to whose careful measurements they are indebted for many of the results which are embodied in this communication .

